As it was mentioned before


It might be a good idea to check in a testing environment the code you trying to deploy. I am assuming that it's not an option in this case. However, would you be able to do it one by one at least, in order to minimise the impact?

Check the bug toolkit, so you know what to expect and where it might break

http://tools.cisco.com/Support/BugToolKit/

Normally that's how I'd do it- but this has been sprung on me without much notice, so I can't...

Upgrade is good and all but take the time and do it right and in steps, I know boss ask and boss gets, but try to tell them that shit can go the wrong way because you had no time to plan this right. Not sure how many devices you have, still 2 days for all out OS upgrade is a push IMO even if you have your network map printed at the back of your skull.

Best of luck!

Hmm, I work in healthcare too.
Just be careful, try and take your time making sure you have a rollback plan if things go wrong. The 4500 is a different beast to the 3560 / 3750's as they are chassis based and have supervisors. The upgrade is similar but you need to take into account the storage location i.e. flash card, supbootflash and whether or not there are dual supervisors.

Like mlan says, 'dir bootflash:' should work, also 'dir slavebootflash:' will show you the standby sup flash if you have one.. or 'dir ?' for anything else i.e. flash disks...

DanC: I don't work in healthcare. I used to work on a global mesh VPN project for literally a full.global.mesh. Yeah, bet you can figure out who paid for that...

I now work in a transportation facility. In some ways much more laid back, in other ways much less. I like it here

The switches I am pretty confident I can handle- after all, we don't have to "jump trains" to the newer IOS. IE: we stay on the same basic version, just going for the latter refinements.

This business about sup engines and so on, I am reticent about.

we used to do 800+ devices in a 12 hour maintenance window.
not just switches and routers, voice, wireless, video gear + linux, too..

_________________
"If you're good at anticipating the human mind. It leaves nothing to chance."
-Jigsaw

Sure thing for the 12h change window, but from "Here's the project" to "done" in 2 days is a bit crazy I think. Perhaps I'm just not super quick thinking.

Agent Zed:

I think the point is- he's talking "we" useed to do this. How many people is "we"? Plus, knowing ristau, he worked with highly-capable folks, since he is one.

In my case, it's me. Noone else. And to be honest with you, I am not highly capable at this stuff... yet.

I mean, FFS, I had to be told how to find out how much memory's on the darn things!

there was also _alot_ of scripting involved,

each device shouldn't take more then 15-20 minutes to load the image and reboot.
staging helps early on, specifically routers, switches no so, if they are routing, the archive
command steals too many cpu cycles and the neighbor adjacencies drop (as least in my past experience)

you can also use the 'reload in' or 'reload at' commands to schedule reloads.

make sure you schedule properly, for example I was starting the archive upgrade
on the WAN switch , then started a archive upgrade on remote switch, when the
wan switch reloaded, the remote switch lost it's file transfer and I had to restart it.

you will also need to verify your versions after upgrade, check logs, and verify everything
is working properly, to make sure all is good before going to bed for the morning.

_________________
"If you're good at anticipating the human mind. It leaves nothing to chance."
-Jigsaw

Everything I am responsible for is within 15 minutes of my location, so I don't think I am going to bother with the archive command or scripting- I've not dealt with that yet.

the archive command is easier than manual upgrade for switches, it will do everything you need, include reload if desired.
but you may run into problems if the .bin files are not in the expected locations for the archive to run, this will cause out of memory errors. so if you've done things manually in the past, stick with that.

_________________
"If you're good at anticipating the human mind. It leaves nothing to chance."
-Jigsaw

Keep in mind that if you have stacked switches and manually upgrade, you have to manually copy the file to each switch's Flash, and do the 'boot system' command a little different.

_________________
"A problem well stated is a problem half solved". (Charles Kettering)

by "stacked switches" I assume you mean, switches with cascade cables between them, aka (I think) clusters?

No, none of that here.

Grr, this is NOT the time for problems...

I've done 1 switch and tried 3 more. First went fine, like this:

hop into the switch via console. Ethernet patched between switch and laptop, too.

dir flash: (check and make sure there's enough room for the .bin. The .bin I am using is 16meg (15,965,969 bytes). I have about 17,500,000 free on the flash)
copy tftp flash:
(copy process starts- I am using a proven TFTP server on the laptop, made sure it was started, etc)
a couple of copy errors show in the TFTP server log... but it claims to complete successfully)
I compared the filesize manually- it's the same as it should be- 15,965,969 bytes.
then do a "verify flash:(filename)". This fails. "File system hash verification failed for (filename). No such file or directory"...

but I can SEE the damn thing, RTF there... and am understandably reticent to continue... Thoughts?

I have done some research and honestly believe this to be a bug.

Feeling courageous (and still being at work some 13 hours after my day started, and everyone else went home 4 hours ago), I have the "least damaging if things go TU") switch proceeding with booting off of the "corrupt" load.

... and while that's spinning, I'm gonna go research restoring a cisco switch from a bad IOS upgrade... :yikers:

Do you still have the original IOS on there? If so the recovery isn't too bad. Just boot to ROMMON and tell it to boot with the other image. If the only IOS on the flash is bad it can be a pain. You have to xmodem a new one up, and it can take a few hours.

-Otanx

HUMM... waiting to hear how this saga turned out....

I hope this wasn't the season finale and
I don't have to wait until next season....

_________________
"If you're good at anticipating the human mind. It leaves nothing to chance."
-Jigsaw

I was right.

Followed process right past the "warning! Warning Will Robinson" crap that the verify command brought up, made the machine boot off the new IOS. All golden. I did 3 more switches afterwards. All is good.

... 9 switches to go, then I have some routers. Did I mention I've never done an upgrade on a router?

easy peazy, just gotta hope everything comes back up afterwards.


copy the ios,
verify the image
clear the old boot var
set the new boot var
reload

_________________
"If you're good at anticipating the human mind. It leaves nothing to chance."
-Jigsaw

and don't forget to 'wr mem' before you reload

_________________
"A problem well stated is a problem half solved". (Charles Kettering)